Variations in the ratio of working-memory to reference-memory trials reveal sample-encoding effects in pigeon short-term memory

Abstract

Two groups of pigeons received training on a delayed conditional discrimination [working-memory (WM) trials] and a simple simultaneous discrimination [reference-memory (RM) trials]. For the same-sample group, the nominal sample stimuli on RM trials were the same as the sample stimuli on WM trials. For the different-sample group, they were different. In Experiment 1, the ratio of WM to RM trials (1:5,1:1, or 5:1) and the delay interval (1, 4, or 8 sec) were manipulated. WM trial accuracy decreased as the ratio of WM to RM trials increased for the different-sample group, but variations in this ratio had little effect in the same-sample group. In Experiment 2, variations in the ratio of WM trials to either RM trials or unfilled time intervals (UTIs) approximately equivalent to the duration of an RM trial were studied. When WM trials occurred in the context of RM trials, variations in the trial ratios produced the same results as in Experiment 1. Substituting UTIs for RM trials had a much greater effect on the performance of the same-sample group than on that of the different-sample group. In the same-sample group, WM trial accuracy was affected by variations in the ratio of WM trials to UTIs but not by variations in WM-to-RM trial ratios. These effects appear to be mediated by changes in sample-encoding processes, rather than by intrusions from previous choices or previous samples.